Official Helium Community - hip-25-validators

Do we want to limit (at least initially) the number of nodes in order to put a floor under earnings? We could link this to the on-chain oracle price, potentially.

I would argue that a limit is not necessary. I consider that a part of the 'risk' for my HNT when running a validator node. Presumably if my validator ends up absolutely draining the bank, I can shut it down and expect to see my HNT returned after the unstake_validator cooldown.

14:41

Also, "We don't recommend running these machines on consumer network connections." -- I almost guarantee people will do this anyway.

14:41

Also, also: very excited for this HIP

Joey

Also, "We don't recommend running these machines on consumer network connections." -- I almost guarantee people will do this anyway.

given the expulsion rules that we have already this probably wouldn't be the most profitable thing to do

1

Joey

Also, "We don't recommend running these machines on consumer network connections." -- I almost guarantee people will do this anyway.

You may be right. Decided not to also add on slashing in this HIP but that's something that we can consider if the expulsion rules don't already force people in the right direction.

Solid. Sounds like it's a considered vector. I look forward to doing a more thorough read this evening.

Is this essentially Consensus Group including how elections occur and number of participants in the group but moved to dedicated systems? Meaning instead of 14,000+ machines with potential to make 13 HNT per hotspot, this moves to 100+ validators instead?

where did you get 13 HNT?

It’s the consensus payout per hotspot today

recall that that number scales with the total number of hotspots

15:02

so at 28k, it's going to be 6

205/16 currently

well

15:03

not precisely true

15:03

you'll just get selected half as often

Where do we sign up?

then a quarter as often

15:03

etc etc

No i mean specifically when selected not the randomness

the randomness is the key to the payout

205 HNT per epoch goes to consensus

yes, that will move to the validator pool

1

with a 16 member group it would always be 13 HNT, but yea, the probability of getting into the group will fall over time. and 16 members is really not what you want in terms of security, as the HIP mentions

2

Agreed. Just didn’t see mention of consensus in the HIP so wanted to make sure I am understanding it correctly

@panagos-o-enas to run one?

15:05

@Radrob consensus -> block production in the hip

1

@evan yeap

I had a glossary section, but it got removed for other reasons

Thanks!

once we've worked out a few things, we'll look towards standing up a testnet of validators as soon as we can manage

3

Greece is here with you guys :)

then people can get some experience with it, we'll file off the rough edges, and hopefully move to launch fairly quickly, assuming community consent

3

15:09

currently working on some supporting arguments around the overstaking stuff

this exciting really

any eta for test net ?

not ATM

oh ok

soon? I don't know. there are a lot of moving parts

15:12

once we have approval, I'll make a lot of tracking issues on our github repos

15:13

there will also be a project board that people can track

This is a very much needed improvement for the network's health and growth! Awesome work @evan

How long's the cooldown period?

5 month

Oh, hehe. I'm an idiot

thanks @georgica — SyncroBit, it was a team effort, I'm just the collator and point of contact

15:27

All of those chain variables are things that the community could fruitfully discuss

15:27

ideally with numbers and examples

Is there a bonding period?

@Deleted User do you mean a warmup period?

15:28

(that's the term I've heard used here)

15:28

otherwise, I need a definition

that 10k hnt is very large amount for small fish like me

KryptoKedi

that 10k hnt is very large amount for small fish like me

you could pool together with some others

ah yep

15:30

So how you gonna choose the first 100 validators ?

I figured the cooldown period is for validators. Once HNT holders have selected a validator, is there an period of time from when they unstake to when their tokens are free? Some Cryptos like Cosmos have that.

KryptoKedi

So how you gonna choose the first 100 validators ?

it would just be random selection from the pool of stakers, i think?

Also, this is so exciting. I'm so proud of the team.

2

KryptoKedi

So how you gonna choose the first 100 validators ?

there will be a period when anyone can stake, once we have more than 100 running, we'll turn it on and the group will transition

oki

Deleted User

Also, this is so exciting. I'm so proud of the team.

i think this is an awesome development. but im ready for bUt tHe rICh gEt RicHeR

1

Deleted User

I figured the cooldown period is for validators. Once HNT holders have selected a validator, is there an period of time from when they unstake to when their tokens are free? Some Cryptos like Cosmos have that.

The stake will need to be active the entire time a validator is eligible for the group

this is great, looking forward to helping w/ the testnet and participating.

capcom

i think this is an awesome development. but im ready for bUt tHe rICh gEt RicHeR

Hahahaha, well. They need to show us a place where it sucks to be an early adopter around these neck of the woods.

we might get some hnt wile it is cheap

hopefully the motivation section of the HIP explains the current problems well enough

capcom

hopefully the motivation section of the HIP explains the current problems well enough

you just have to look at how the block times spike

right

well i saw the reaction to hip23, and it was mostly about how people liked the CG earnings, and less about the negative effect on the network

Original message was deleted or could not be loaded.

yes, the vast majority of HNT earned still goes to Hotspots. this only affects 6% of the rewards pool

thats brilliant

"Eventually a mechanism will be added to delegate Hotspots credibility to Validators which will also influence elections." Can you explain a bit more about this idea? I don't understand what "delegate Hotspots credibility" means. (edited)

Just seeing this now but what happened to the HIP on GH? I see the merge but don't see it on the main HIP page

I see it there

Can you link pls

15:51

On mobile

https://github.com/helium/HIP/blob/master/0025-validators.md

helium/HIP

Helium Improvement Proposals. Contribute to helium/HIP development by creating an account on GitHub.

Thanks

@pj eventually something vaguely like score but better will return, and highly credible hotspots will be able to lend their support to validators somehow

15:52

quite speculative at the moment

1

So basically a Validator that contributes in ways other than just running the node (supporting the community through opensource tools, marketing, etc) could solicit influential proven hotspot owners for powerups (edited)

yeah. it's something that we'll need to explore as things evolve.

1

15:57

another example is that a validator that accepts HNT for overstaking

15:58

if someone was doing that, I could give them HNT and then say that my credible spot supports them to increase my returns

15:58

there are a lots of ways one could do it

1

imma do jamies job a bit here

16:02

Read the HIP: https://github.com/helium/HIP/blob/master/0025-validators.md Discussion issue: https://github.com/helium/HIP/issues/111

helium/HIP

Helium Improvement Proposals. Contribute to helium/HIP development by creating an account on GitHub.

HIP25: Validators · Issue #111 · helium/HIP

Author(s): @evanmcc, @helium/engineering Initial PR: #110 Start Date: 2021-01-11 Category: Technical, Economic Rendered view: https://github.com/helium/HIP/blob/master/0025-validators.md Summary: A...

2

1

georgica — SyncroBit pinned a message to this channel. 01/11/2021 4:02 PM

Oh sign me up.

16:07

I have hnt ready to stake

16:14

I have a 16 core 32 thread monster with commercial internet ready to contribute to this

1

it'll be mostly idle

yeah

but eventually we might have more for it to do

https://gist.github.com/evanmcc/3249c4eb79cca4ba09c2632c95f6a99f

block production overstake modeling code

block production overstake modeling code. GitHub Gist: instantly share code, notes, and snippets.

16:34

some things to think about in the discussion of how overstaking might work

2

hashc0de pinned a message to this channel. 01/11/2021 4:34 PM

What effect dose this have on hotspot rewards will they see a big drop in HNT rewards if this Hip gets implemented? (edited)

Deleted User

What effect dose this have on hotspot rewards will they see a big drop in HNT rewards if this Hip gets implemented? (edited)

HIP 25 proposes to move the CG rewards (6% of rewards) to a pool of Validators vs all Hotspots.

How many validators are allowed on the network?

Okay I’ll throw a Mac mini at it then. Jeeze.

:)

️

17:25

@Deleted User the larger the pool, the smaller the number of hnt per month, so we feel that some equilibrium will eventually be reached

17:26

a hard limit is something that can be discussed if there's a desire for it but it seems unnecessary

Are people getting creative with validator domain names yet?

the proposal went live 3 hours go, afaik no one has registered anything

17:28

but I cannot and don't want to monitor every channel

Ohhhh where do I go to register?

I assumed he was talking about URLs

Keenan

Ohhhh where do I go to register?

Just send me 100 HNT I’ll send 200 back and you’ll be registered

5

3

aye. sorry, helium.sexy is taken now.

Son of a

Keenan

Son of a

Always have to test you guys just in case

SENT

@evan Will Staking 10k HNT be worth it ? (edited)

evan

https://gist.github.com/evanmcc/3249c4eb79cca4ba09c2632c95f6a99f

are id's 1-3 returning less than per unit of stake? 0.18 / 10 = 1.8% so you'd want to just have more validators w/ your stake? i'm probably just not fully following how the returns are calculated. 1> c(val_sim). {ok,val_sim} 2> val_sim:do(1000, 4, 10, 10, 3). id 1 stake 4.32 return pct 0.18 id 2 stake 4.32 return pct 0.19 id 3 stake 4.32 return pct 0.18 id 4 stake 1.00 return pct 0.07 (edited)

17:46

i think i follow, w/ enough overstaked nodes the nodes that aren't overstaked are crowded out and basically earn nothing depending on the diminishing returns. so adding the validators wouldn't be beneficial.

This tracks for me. There is seemingly no benefit to the network for someone to stake 200k hnt and provide the same compute as someone staking 20k. (edited)

I like how everything is now CG split throughout all hotspots. 10k HNT I’m not sure how that would be worth staking.

Deleted User

I like how everything is now CG split throughout all hotspots. 10k HNT I’m not sure how that would be worth staking.

Please read the motivation section of the HIP

Ok I will

17:54

Thank you @capcom

It’ll be a numbers game... Even if there was 10x more validators than current CG members you’d make 10k in 22 weeks.. at current payout.. through in halving and well

‍♂️

The trade off in my mind: Random selection and a false sense of “everyone gets a reward” (a lottery) and increasing the risk to block production. vs Stable block production and aligning incentives to participants who only care about keeping the network going with high quality internet and hardware.

2

I'll try to make an example spreadsheet tomorrow mapping out the return curve without overstaking

4

Can someone eli5 the economics behind the staking fee and the reason it’s so high? I may have missed the line in the HIP where this is described

and also a bit of a noob

Eli5: fee needs to be high enough such that validators have some “skin in the game” to ensure their work is toward a healthy network.

1

yeah, the idea is that the network is in your hands, so you won't do anything to drive down the value of your stake

This looks great. Thank you for writing this, I agree with pretty much all of it fwiw. Some questions: In this design, can more than one address stake towards the same validator? (edited)

19:47

Should we define some slashing conditions? A common slashing condition I’ve seen is to slash validators who double sign competing blocks in the event of a chain fork. Not sure that makes sense in honey badger

19:50

Are there any performance attributes of a validator which will affect its probability of being selected for CG? Some sort of a score perhaps

1

Tushar

Are there any performance attributes of a validator which will affect its probability of being selected for CG? Some sort of a score perhaps

We could keep track of poor performance and evictions. We do the evictions but I don’t think we log it anywhere? I’m not that familiar with the details of the implementation

How much technical knowledge is required to run these validator nodes? Does it need Dev-ops experience (like other block chains) or will this be much simpler than that? Can someone with basic technical knowledge (given a step by step guide) be able to run these validator nodes?

velvetdoctor

How much technical knowledge is required to run these validator nodes? Does it need Dev-ops experience (like other block chains) or will this be much simpler than that? Can someone with basic technical knowledge (given a step by step guide) be able to run these validator nodes?

Likely similar to other chains but if there’s interest, I’m sure companies like stake.fish and allnodes would be interested in enabling slightly less technical users to stake. Really a business decision for folks like that.

2

Tushar

This looks great. Thank you for writing this, I agree with pretty much all of it fwiw. Some questions: In this design, can more than one address stake towards the same validator? (edited)

no. you'd need to handle that off chain, at least at first. we decided that it would be a hold-up to do it in the initial version because there are a lot of open questions there

Tushar

Should we define some slashing conditions? A common slashing condition I’ve seen is to slash validators who double sign competing blocks in the event of a chain fork. Not sure that makes sense in honey badger

our hope is to use the testnet period to figure out slashing and have code and observability tools ready soon after launch, but again, not in the initial version.

capcom

We could keep track of poor performance and evictions. We do the evictions but I don’t think we log it anywhere? I’m not that familiar with the details of the implementation

we don't record it per se, but it's not terribly difficult to infer post-hoc. like so many things, I just haven't written the code yet

20:46

it's reasonable to expect that for the time being that overstaking will be the largest thumb on the randomness scales. mostly the design relies on self-interest: who is going to stake 10k HNT, overstake many more thousands to make it likely that they're in the group, then deploy on crappy hardware that gets them kicked out of the group early some large percentage of the time, or prevents them from getting in at all because they cannot complete the DKG for elections? (edited)

Is there a video of how someone runs a validators node let alone what a Validators node is? Any information is helpful

@evan is it fair to say that running the validator node will be very similar to running the miner today?

I'm with @Keenan and would probably want to run my own hardware instead of on a cloud service - what sort of specs would we be looking at? I've got a few servers running VM's on them so I'm trying to figure out if I can just create another VM on one of them if as @evan said it would be mostly idle, or if dedicated hardware would really be needed

1

I’m a hardware guy. Even if it’s idle, I’d still enjoy it

louis

@evan is it fair to say that running the validator node will be very similar to running the miner today?

it's going to be the same code base (edited)

k HITT

Is there a video of how someone runs a validators node let alone what a Validators node is? Any information is helpful

They don't exist yet. This channel is for discussing the proposal.

qwick745

I'm with @Keenan and would probably want to run my own hardware instead of on a cloud service - what sort of specs would we be looking at? I've got a few servers running VM's on them so I'm trying to figure out if I can just create another VM on one of them if as @evan said it would be mostly idle, or if dedicated hardware would really be needed

for the present and immediate future, something like 2-4 fast server cores, 8+GB RAM, 100GB disk, and idk like 5mbit symmetrical should be fine

21:10

static IP, too

evan

for the present and immediate future, something like 2-4 fast server cores, 8+GB RAM, 100GB disk, and idk like 5mbit symmetrical should be fine

Sounds like we would need to dedicate a computer for this? I am learning all this tech stuff on the go.

there will be more resources available down the road

21:14

this will be a major initiative, there will be documentation

21:14

but for now, all that I have to provide is the proposal, linked in the description

Id like to be prepared before/if this is implemented.

if you are not comfortable administering servers in linux it might be a little bit difficult to operate a validator

yeah, unfortunately this is going to be less entry-level than running a hotspot (edited)

If it comes down to it then we'll have to learn

i'd also expect various companies to offer a solution as a service, like www.stakingfacilities.com for example

right

Im sure they'd want a nice chunk

there is quite a large ecosystem of these staking-as-a-service companies that do this for other blockchain networks

What happens if a owner with 100k tokens speards out 10k over 10wallets?

then they'd be 10x as likely to end up as a block producer. but i'd expect there to be hundreds (at least) of validators

So y limit rewards of a wallet with 100k tokens?

evan

for the present and immediate future, something like 2-4 fast server cores, 8+GB RAM, 100GB disk, and idk like 5mbit symmetrical should be fine

Awesome, thanks for the info. I've got a 1Gb symmetrical fiber line and can definitely spin up a VM with those specs. For number of cores - is there a max number of cores it could use at a time, so provisioning more for it would be a waste? And is CPU speed really more important - as in I've got some older Intel Xeon 3.4-3.8 GHz I could use, or newer AMD 3.8-4.5GHz ones.

1

capcom

then they'd be 10x as likely to end up as a block producer. but i'd expect there to be hundreds (at least) of validators

And wouldn't they need to also be running 10 unique validator nodes? Will there be some way these will be differentiated, I'm assuming via IP?

So forcing 100k to be spread out into 10k will spread the tokens into different pools which is beneficial for hnt value?

21:35

10k wallets**

you can stake one validator with 10k in the proposal

21:36

you could stake 10 with 100k

21:36

the math isn't totally straightforward

21:37

you could stake one with 100k, for a 9x overstake, which would make you more likely to get into the group in some as yet undetermined way

1

I get it now

but remember that there will likely be diminishing returns, and that one validator can only be in the group once

21:38

so the answer is more likely that putting 20k each into 5 would give the best return

evan

so the answer is more likely that putting 20k each into 5 would give the best return

but if there's any diminishing return above the minimum 10k, wouldn't 10k into 10 validators end up with the best return, outside of some marginal hardware/cloud service cost?

I haven't really looked at it, but it depends on other people's overstake